Accelerated Vector Pruning for Optimal POMDP Solvers
Authors: Erwin Walraven, Matthijs Spaan
AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments show that our algorithm improves the performance of existing pruning methods, and our results show that the accelerated pruning algorithm creates the fastest variant of incremental pruning for POMDPs. |
| Researcher Affiliation | Academia | Erwin Walraven, Matthijs T. J. Spaan Delft University of Technology Mekelweg 4, 2628 CD Delft, The Netherlands |
| Pseudocode | Yes | Algorithm 1: Vector pruning (White & Lark); Algorithm 2: Find Belief Std computes the belief in which w improves U the most; Algorithm 3: Find Belief Dec computes the belief in which w improves U the most |
| Open Source Code | No | The paper does not provide a link to its own open-source code or state that its code is publicly available. |
| Open Datasets | No | The paper mentions using 'all POMDP domains used by Cassandra, Littman, and Zhang (1997), Feng and Zilberstein (2004) and Raphael and Shani (2012)', implying standard benchmarks, but does not provide direct access information (e.g., URL, DOI, or a proper citation with author/year in brackets for dataset access) for these datasets. |
| Dataset Splits | No | The paper does not specify training, validation, or test dataset splits. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions using 'the LP solver GLPK' and that 'Results for Gurobi and lpsolve can be found in the supplement', but does not specify version numbers for these software components. |
| Experiment Setup | Yes | For each domain we consider the first 30000 LPs that are solved during the execution of incremental pruning (or until the problem is solved or memory limits are exceeded, details in the supplement). For each LP we execute the standard LP and the decomposed LP, during which we measure the running times and the number of constraints added by the decomposed LP. |